Sunlight-driven water-splitting using two dimensional carbon based semiconductors

Sunlight-driven water-splitting using two-
dimensional carbon based semiconductors
Pawan Kumar, a
Rabah Boukherroub *b
and Karthik Shankar *a
The overwhelming challenge of depleting fossil fuels and anthropogenic carbon emissions has driven research
into alternative clean sources of energy. To achieve the goal of a carbon neutral economy, the harvesting of
sunlight by using photocatalysts to split water into hydrogen and oxygen is an expedient approach to fulfill
the energy demand in a sustainable way along with reducing the emission of greenhouse gases. Even though
the past few decades have witnessed intensive research into inorganic semiconductor photocatalysts, their
quantum efficiencies for hydrogen production from visible photons remain too low for the large scale
deployment of this technology. Visible light absorption and efficient charge separation are two key necessary
conditions for achieving the scalable production of hydrogen from water. Two-dimensional carbon based
nanoscale materials such as graphene oxide, reduced graphene oxide, carbon nitride, modified 2D carbon
frameworks and their composites have emerged as potential photocatalysts due to their astonishing
properties such as superior charge transport, tunable energy levels and bandgaps, visible light absorption,
high surface area, easy processability, quantum confinement effects, and high photocatalytic quantum yields.
The feasibility of structural and chemical modification to optimize visible light absorption and charge
separation makes carbonaceous semiconductors promising candidates to convert solar energy into chemical
energy. In the present review, we have summarized the recent advances in 2D carbonaceous photocatalysts
with respect to physicochemical and photochemical tuning for solar light mediated hydrogen evolution.
Dr Pawan Kumar received his
PhD degree in Chemical Science
from CSIR-Indian Institute of
Petroleum, Dehradun, India in
2016 under the guidance of Dr
Suman L. Jain. Prior to PhD, he
worked at R&DE, DRDO, Ministry
of Defense, Pune, India, as
a Junior Research Fellow. During
his PhD, he was awarded the
prestigious Raman-Charpak
fellowship in 2014 by Campus
France to work with Prof. Rabah
Boukherroub (IEMN & University of Lille, France). Aer his PhD, he
joined the research group of Prof. Radek Zboril at RCPTM, Palacky
University, Czech Republic as a postdoctoral fellow. Currently, he is
a postdoctoral researcher at the University of Alberta, Canada in the
group of Prof. Karthik Shankar. His research interests are focused on
the synthesis of advanced photocatalytic materials for energy appli-
cations, the heterogenization of molecular catalysts, the modication
of carbon nitride and the development of photocatalysts for CO2
activation, water splitting and photo-organic transformations.
Dr Rabah Boukherroub received
a PhD in chemistry from the
University Paul Sabatier in
Toulouse, France. He is
currently a CNRS research
director and a group leader at
the Institute of Electronics,
Microelectronics and Nanotech-
nology (IEMN), University of
Lille, France. He is an Associate
Editor for ACS Applied Materials
& Interfaces. He is also a guest
Professor, China University of
Petroleum, Qingdao, China. His research interests are in the area
of nanostructured functional materials, surface chemistry, and
photophysics of semiconductor/metal nanostructures with
emphasis on biosensors, nanomedicine, photocatalysis and energy-
related applications. He is a co-author of 480+ research publica-
tions and wrote 31 book chapters on subjects related to nano-
technology, materials chemistry, and biosensors. He has 11
patents or patents pending.
a
Department of Electrical and Computer Engineering, University of Alberta,
9211-116 St, Edmonton, Alberta, Canada T6G 1H9. E-mail: kshankar@ualberta.ca
b
Univ. Lille, CNRS, Centrale Lille, ISEN, Univ. Valenciennes, UMR 8520 – IEMN, F-
59000 Lille, France. E-mail: rabah.boukherroub@iemn.univ-lille1.fr
Cite this: J. Mater. Chem. A, 2018, 6,
12876
Received 4th March 2018
Accepted 23rd May 2018
DOI: 10.1039/c8ta02061b
rsc.li/materials-a
12876 | J. Mater. Chem. A, 2018, 6, 12876–12931 This journal is © The Royal Society of Chemistry 2018
Journal of
Materials Chemistry A
REVIEW
Publishedon24May2018.DownloadedbyUniversityofAlbertaon8/2/20189:24:30PM.
View Article Online
View Journal | View Issue

1. Introduction
The two biggest long term challenges of modern technological
civilization are global warming due to anthropogenic carbon
emissions and the steadily rising global energy demand. The
problem of shortage of fossil fuels and increased CO2 concen-
tration in the environment due to burning of fossil fuels bring
up the urgency to search for carbon neutral clean energy to
sustain environmental and energy issues.1
Due to the expo-
nential growth of world population and global economic
growth, the world's energy consumption is expected to double
by 2050 and triple by the end of the century relative to the
present consumption (14 TW).2
Sunlight is an inexhaustible and
continuous source of energy; approximately 4.3 Â 1020
J of
energy strikes the surface of the earth per hour, which is more
than the current consumption of energy by all of humanity in
one year.3,4
Plants and phytoplankton are able to harvest this
energy and approximately 100 Terawatts of energy is harvested,
which is about six times larger than the power consumption of
human civilization.5
Plants convert solar energy into chemical
energy by two reactions: (i) splitting of water into hydrogen and
oxygen in which hydrogen is trapped as reduced nicotinamide
adenine dinucleotide phosphate (NADPH), while oxygen is
released into the environment to be used by other creatures for
food oxidation, and (ii) conversion of CO2 into organic mole-
cules using electrons derived from water splitting. Nature has
thus demonstrated that solar energy can be stored in the form
of chemical bonds, which can be used to meet our energy
demand.6
Although photovoltaic devices have been deployed to
convert solar energy into electrical energy with up to 43.5%
efficiency, these devices encounter certain disadvantages such
as intermittency, high production costs and long energy
payback times.7,8
Hydrogen has been considered as a clean solar fuel due to its
high free energy content (DG ¼ +237 kJ molÀ1
), easily accessible
water resource, and production of water and oxygen as sole by-
products.9,10
Alternatively, hydrogen due to its high energy
content can convert CO2 into hydrocarbons such as methanol,
which is compatible with internal combustion engines. Addi-
tionally, as a liquid phase, it can be stored and transported
easily. Currently, hydrogen is produced on a large scale by
reforming (dry or steam reforming) of hydrocarbons or through
gasication of fossil fuels, both of which produce large amounts
of CO2.11
The photocatalytic production of hydrogen from water
splitting is a viable approach to sustain the process which relies
on two most abundant resources, water and sunlight.12
Solar
light mediated water splitting can be achieved by two methods:
(i) using a photoelectrochemical cell where a semiconductor
electrode, immersed in water, is irradiated with light and
another electrode works as a carrier counter electrode; (ii) using
a photocatalytic reactor wherein particles of a semiconductor
photocatalyst suspended in water act as micro-photoelectrodes
and facilitate oxidation and reduction reactions of water on
their surface.
Since the initial experiment by Fujishima and Honda in 1972
on photoelectrochemical water splitting using a TiO2 photo-
electrode,13
numerous semiconductor materials relying on d-
block elements have been explored for solar light mediated
hydrogen production. Subsequently, Bard's group demon-
strated photocatalytic water splitting using suspended semi-
conductor particles.14
Both photoelectrochemical and
photocatalytic methods have pros and cons. Photocatalytic
systems are extremely simple and have lower operating costs,
but cannot prot from the use of a bias to separate charge
carriers, thus limiting the achievable efficiency of conversion of
sunlight into chemical energy; also, the separation of evolved
hydrogen and oxygen poses a problem.15
Up until now, different
material systems such as oxides, suldes, phosphides, nitrides,
oxynitrides, oxysuldes, etc., have been developed for the pho-
tocatalytic conversion of solar energy into chemical energy.16
However, even aer 45 years of research, the quantum efficiency
remains too low to justify the scaling up of hydrogen production
to the industrial level. An efficient photocatalytic system which
can provide four consecutive proton coupled electron transfer
(PCET) steps to achieve overall water splitting with a high
photoconversion efficiency is still a challenge to the scientic
community. Overall water splitting involves two half reactions –
the oxidation of water which produces O2, protons and elec-
trons (OER), and the reduction of protons which produces
hydrogen (HER), eqn (1)–(3). In order to sustain the process of
water splitting, both OER and HER processes should occur
simultaneously to chain the ow of electrons.
Overall water splitting
2H2O + hv / 2H2 + O2 (1)
Oxygen evolution reaction (OER)
2H2O / 4H+
+ 4eÀ
+ O2 (2)
Dr Karthik Shankar is an Asso-
ciate Professor in the Department
of Electrical & Computer Engi-
neering at the University of
Alberta. He received M.S. and
PhD degrees in Electrical Engi-
neering from The Pennsylvania
State University, and obtained
his B.Tech degree from the
Department of Metallurgy and
Materials Science at the Indian
Institute of Technology-Madras.
He is an Associate Editor for
IEEE Sensors Letters and an Editor for the Journal of Materials
Science: Materials in Electronics (JMSE). He joined the University of
Alberta in 2009 and received the Petro-Canada Young Innovator
Award in 2010. From 2012–2017, he held a secondment to NRC-
NINT as a Research Officer. His research interests consist of
organic semiconductors, nanomaterials, plasmonics, solar cells and
photocatalysts. As of April 2018, he has authored more than 100
journal articles that have collectively been cited over 16 000 times.
This journal is © The Royal Society of Chemistry 2018 J. Mater. Chem. A, 2018, 6, 12876–12931 | 12877
Review Journal of Materials Chemistry A
View Article Online

Hydrogen evolution reaction (HER)
4H+
+ 4eÀ
/ 2H2 (3)
The reduction potential of H+
/H2 is 0.00 V at pH 0 vs. NHE;
however À0.41 V overpotential is needed for the production of
hydrogen in aqueous solutions at pH 7 vs. NHE, while the
potential for water oxidation H2O/O2 is +1.23 V vs. NHE at pH
0 (or +0.82 V vs. NHE at pH 7). According to these values,
a minimum semiconductor band gap energy of 1.23 eV is
required to drive the overall water splitting reaction. In practice,
a slightly larger bandgap of 1.35 eV or higher is required to split
water due to overpotential-related losses. The value of potential
can be changed to any pH by using equation E0
¼ E0
(pH0) À
0.06pH, where E0
is the standard reduction potential.17,18
Furthermore, the position of the semiconductor valence band
should be more positive than +0.82 V vs. NHE at pH 7, while the
position of the conduction band should be more negative than
À0.41 V vs. NHE at pH 7.
Many inorganic semiconductors such as CoP, NaTaO3, ZnO,
InVO4, ZnGa2O4, ZnS, Ta3N5, Cu2ZnSnS4, and CdS and mixed
metal oxides such as RuxTi1ÀxO2, Rh–Cr mixed oxide nano-
particles, dispersed (Ga1ÀxZnx)N1ÀxOx and ZrO2-supported Nix-
Fe3ÀxO4, etc., are developed as water splitting
photocatalysts.20–22
However, solar to fuel conversion efficiency
is still limited to 1–2% due to charge recombination, lack of
visible light absorption, and poor charge separation. A low
electronic band gap to absorb in the visible region, appropri-
ately long carrier lifetimes and/or carrier diffusion lengths,
earth abundance, high stability, nontoxic nature and easy
processability are the prime requirements of an ideal photo-
catalyst. Unfortunately, many promising semiconductors,
especially oxides, have a wide band gap and absorb light below
400 nm, which represents only 4% of the whole solar spectrum
(Fig. 1). If all the incident UV light on Earth up to 400 nm can be
harvested, only 2% solar conversion efficiency can be achieved,
which is equal to the maximum conversion efficiency of
photosynthesis in green plants (1–2%).23
Furthermore, the
inherent health hazards of UV light require specially designed
reactors with quartz windows, which add an extra cost to the
process. So, in order to achieve maximum solar conversion
efficiency, it is of utmost importance to harvest visible light
which is approximately 45% of the total solar spectrum. If it is
possible to harvest solar light up to 600 nm, the conversion
efficiency can be increased up to 16% which can be further
increased up to 32% if light up to 800 nm can be captured.
Furthermore, charge recombination is prevalent and it is esti-
mated that only 10% of photogenerated electron–hole pairs are
available for water splitting. Charge recombination can take
place on the surface or in the bulk of semiconductors (Fig. 2a).
Surface recombination can be slowed down by forming heter-
ojunctions through surface decoration with noble metal nano-
particles, which causes the formation of Schottky barriers;24,25
the built-in eld associated with the Schottky barrier prevents
fast recombination. Another option is to use promoters – hole
capturing agents, i.e., IrO2, CoOx, RuO2, etc.,26–28
(Fig. 2b) or
perform the catalytic reaction in the presence of hole scavengers
such as alcohols. Volume recombination can be prevented by
reducing the semiconductor particle size29
or modifying the
surface morphology (nanotubes and nanospikes) so that more
electrons and holes can reach the surface and are available for
the photoreaction. The ideal nanostructure is one where the
processes of light absorption and charge separation are
orthogonalized and wherein the maximum distance that pho-
togenerated carriers have to travel to reach the site of chemical
reaction is smaller than the retrieval length in that material.30–32
Apart from these approaches, the hybridization of a nano-
structured large band gap n-type semiconductor with a low
band gap p-type semiconductor to create a type-II (staggered) p–
n heterojunction is also a useful strategy.34,35
The depletion
region or energy band offset at the heterojunction interface
creates a driving force for efficient charge separation, thus
Fig. 1 Band gap and band edge positions of a range of semiconductors vs. NHE at pH 7 representing the redox potential of water splitting.
Adapted with permission from ref. 19.
Journal of Materials Chemistry A Review
View Article Online

preventing geminate recombination. The low band gap p-type
semiconductor can absorb visible light and transfer the pho-
togenerated electrons to the conduction band of the large band
gap semiconductor, while photogenerated holes in the p-type
react directly with the reactants on the surface (direct mecha-
nism) (Fig. 3). Another more fascinating approach involves
mixing of two low band gap semiconductors one of which has
a more positive valence band (oxidative) and facilitates water
oxidation, while the other has a more negative conduction band
to facilitate proton reduction (reductive). Individually none of
them can sustain overall water splitting, while when they are
mixed together they can sustain the process of overall water
splitting. Following the absorption of light, photogenerated
electrons in the conduction band of the oxidative semi-
conductor annihilate photogenerated holes in the reductive
semiconductor. These electrons in the conduction band of the
reductive semiconductor facilitate proton reduction, while
holes in the valence band of the reductive semiconductor
facilitate water oxidation. This mechanism is called the Z-
scheme mechanism and needs two photons for the comple-
tion of overall water splitting (Fig. 3).36,37
Many examples of Z-
scheme photosystems for water splitting have been described
in the literature, including SrTiO3:La,Rh and BiVO4:Mo,38
BiVO4
and Ru/SrTiO3:Rh,39
Ir/CoOx/Ta3N5 and Ru/SrTiO3:Rh,40
etc.
Generally Z-scheme catalysts require an electron mediator, i.e.,
(Fe3+
/Fe2+
, [Co(bpy)3]3+/2+
, graphene), which facilitates charge
transfer at the interface of contact.41,42
Although a lot of research has been dedicated to reducing
the bandgap of semiconductors to achieve high solar to
hydrogen (STH) conversion, photocatalytic performance is still
limited by low absorption coeﬃcients for visible regime
photons, poor quantum yields, fast carrier recombination,
Fig. 2 (a) Schematic illustration of the charge generation, separation, recombination and conversion processes in photocatalysts and (b)
facilitation of charge separation by electron and hole capturing agents. Adapted with permission from ref. 24 and 33.
Fig. 3 Mechanistic representation of direct and Z-schemes of photocatalysis on two diﬀerent band gap semiconductors for overall water
splitting.
View Article Online

photo-corrosion and reliance on expensive noble metal co-
catalysts. In recent years, two-dimensional (2D) carbon based
materials have emerged as new photocatalytic materials due to
their excellent electronic, optical, physicochemical and surface
properties.43–48
Carbonaceous materials are earth abundant and
provide a niche for nanoscale modication in the chemical and
morphological structure, which facilitates efficient charge
separation.49
In the past decade, the exponential rise in the
number of reports on carbon materials for photocatalytic
applications makes evident their wide potential for various
applications. Numerous carbon based 2D materials such as
graphene, graphene oxide (GO), carbon nitride (C3N4), poly-
imide polymers, holey graphene (C2N), doped carbon quantum
dots and their 2D derivatives have been introduced.50–52
Carbon
materials being composed of conjugated sp2
carbon networks
provide excellent charge carrier mobility on their surface. The
high surface area of 2D materials and rapid charge transport
over the surface of conjugated carbon based materials result in
short paths for carriers to reach reaction sites and remove the
reliance on metal based electron and hole capturing agents thus
providing opportunities to create metal-free photocatalysts.53–55
Among various carbon based materials, graphene is the most
intensively investigated in photocatalytic applications to
increase the performance of inorganic semiconductors due to
its exceptionally conductive surface and high specic surface
area; however, the absence of a band gap restricts its role in
photosensitization.56–59
Although a limited increase in photo-
catalytic performance has been demonstrated for several
graphene/inorganic hybrids, the inorganic semiconductor
remains the charge generation center along with its inherent
drawbacks.60–62
Band gap opening in graphene by doping with
heteroatoms and promoting defects is the most effective
method to convert metallic graphene into a semiconductor to
exploit its sub-nanoscale properties.63–66
The differential elec-
tronic densities and divergent atomic centers give rise to various
active sites for photoreactions. While doped graphene materials
can absorb a wide swathe of photons in sunlight to generate
electron–hole pairs due to their narrow bandgap, the VB
maxima remain unfavorable and produce poorly oxidative holes
to facilitate water oxidation.67–69
The band gap broadening can
be achieved by adding high dopant concentrations, but this
reduces the number of p-conjugated conductive nanochannels
required for efficient charge separation.70
Furthermore, diffi-
culties in the mass production of graphene/doped graphene
limit the application of this frontier material in photocatalysis.
Doped carbon derived from a heteroatom-rich source, pos-
sessing less ordered doped graphenic domains, seems
amenable to scale-up and mass production due to the ease of
synthesis. However, doped carbon inherits the drawbacks of
graphene such as a very narrow band gap, self-degradation, and
deactivation of active sites due to which their application
beyond electrocatalysis is less promising.71–74
On the other
hand, graphene oxide (GO) behaves as a semiconductor pos-
sessing a variable band gap depending on oxygen concentration
due to the presence of defects and oxygen rich sp3
hybridized
carbon domains. The maximum hydrogen yield achieved using
GO was 2833 mmol hÀ1
in the presence of a methanol sacricial
donor in contrast to 46.6 mmol hÀ1
in the absence of a sacricial
donor.75
The formation of hybrids and heterojunctions of
graphene/graphene oxide with inorganic semiconductors/dye
molecules/metals and doping have been used to achieve high
H2 yields in the presence of sacricial reagents.60,76–79
However,
the use of methanol and other sacricial reagents is undesirable
for sustainable hydrogen production. Therefore, an ideal
material should have the requisite band structure and active
sites to facilitate the simultaneous reduction and oxidation of
water instead of requiring the use of sacricial electron or hole
scavengers.
Recently, 2D polymeric semiconductors have been envisaged
as future photocatalysts due to the ease of synthesis from cheap
resources, superior visible light absorption, tunable band gap
and easy processability.48,53,80,81
Graphitic carbon nitride (g-
C3N4), a polymeric material composed of triazine (C3N3) or tri-s-
triazine/heptazine (C6N7) units with alternate C and N,
provoked excitation in the photocatalysis community due to its
resilient stability (chemical inertness, thermal resistance and
absence of photocorrosion), appropriate band structure and
ease of synthesis from benign sources.82–86
Since the rst report
by Wang et al. in 2009 on photocatalytic water splitting using g-
C3N4, this eld has witnessed intensive research activity.87
g-
C3N4 possesses a moderate band gap of 2.6–2.7 eV and its CB
and VB redox potential values lie at À1.1 eV and +1.6 eV,
respectively with a substantial lifetime span of the photoexcited
state (nanosecond regime) making it a suitable photocatalytic
material for water splitting.88
Apart from the band gap, the
presence of numerous nitrogen rich sites and ordered defects
provides active centers for the substrate interaction, which
increases reactant concentration on the catalyst surface for
catalytic and photocatalytic reactions.80,89–91
However, extending
the visible light absorption prole of g-C3N4 beyond 450–
460 nm and increasing the carrier lifetime are the major chal-
lenges. Numerous nanoarchitecture assemblies of g-C3N4 and
inorganic semiconductor/dye molecules/graphenic materials
have been designed for improving charge separation and visible
light absorption parameters.92
Doping with metal atoms/ions
and heteroatoms has also been used to shi the visible light
absorption prole towards longer wavelengths.93–97
The
synthesis of carbon nitride materials from nitrogen rich
precursors affords porous bulk g-C3N4.98
In bulk carbon nitride,
graphenic C3N4 sheets remain stacked together, which
compromises the photocatalytic efficiency due to a substantial
amount of concealed surface that is unexposed to light and
reactants. These stacked sheets can be metamorphized into few
layered 2D C3N4 sheets by various solvent intercalations.48,99–102
However, the exfoliation of g-C3N4 to sheets works to the
detriment of photocatalytic performance due to band gap
broadening and loss of crystallinity due to abundant intralayer
hydrogen bonds between sheets, which promote charge locali-
zation and lead to poor intralayer carrier transport.103–105
Furthermore, residual functionalities also act as carrier trap-
ping centers, which reduce the quantum efficiency of photo-
catalysis. Hence forming few layered g-C3N4 sheets while
maintaining crystallinity to decrease trap sites is the best
approach to synergistically achieve a combination of high
View Article Online

surface area and good charge transport/separation. Actually,
crystalline carbon nanosheets have afforded some of the high-
est reported H2 yields among carbon based materials as high as
9577.6 mmol hÀ1
gÀ1
with a quantum efficiency of 9.01% at
420 nm, which was 15.5 times higher than that of bulk g-
C3N4.106
The nanosheets of carbon nitride can be further
modied with quantum dots, graphene, doped graphene, and
doped carbon nitride sheets to improve photocatalytic
performance.107–109
Although g-C3N4 is a good hydrogen evolution catalyst, its VB
band edge potential is less positive so its water oxidation
performance remains underwhelming. The photocatalytic
performance of g-C3N4 to achieve overall water splitting can be
improved by introducing electron rich or decient units into the
framework, which shi band edge positions such that the band
structure of g-C3N4 can be tuned. The polyimide derivative of
carbon nitride synthesized by heating triazine precursors and
dianhydride precursors at high temperature offers tunable
band gap imide semiconductors.110–112
The band structure of
these polyimide polymers depends on the ratio of triazine/
heptazine and dianhydride, and by controlling the type of
monomer unit, a more oxidative VB position can be achieved.
Apart from band edge tuning, these materials possess a higher
visible light absorption and the presence of different charge
density units facilitates better charge separation within the
sheets.
Certain other 2D semiconductors with better optical and
charge transport properties have been reported in the past few
years. One of these semiconductors is C2N (holey graphene)
whose narrow band gap and ordered porous structure make it
a suitable candidate for various applications, including water
splitting.113
Theoretical investigations suggested that
combining C2N nanohybrids with other 2D semiconductor
materials is a promising approach to optimally harvest sunlight
with a high quantum efficiency.114
2D–2D hybrids are better
than conventional hybrids due to superior face to face interac-
tions, which facilitate better charge transfer and carrier
mobility. However, for C2N hybrids, most of the discussion is
limited to theoretical calculations and more efforts are needed
to experimentally realize such nanohybrids.
This review represents an extended discussion on various 2D
carbon materials, focusing on their synthesis, properties, pho-
tocatalytic performance, and chemical and structural modi-
cation of their framework via various approaches to achieve an
appropriate band structure for water splitting. The initial
sections of this review will cover graphene/graphene oxide
based materials with a specic focus on improved charge carrier
separation and band gap opening via various dopants and their
nanocomposites with inorganic semiconductors for hydrogen
generation. Next, carbon nitride based polymers as semi-
conductors will be surveyed, and improvements in their pho-
tocatalytic performance through surface engineering, 2D sheet
transformation, band structure manipulation and particularly,
chemical modication by replacement and/or doping of triazine
or heptazine units will be reviewed in depth. Later in this
review, 2D polyimide polymers, carbonaceous-inorganic 2D–2D
semiconductors, emerging carbon based semiconductors, i.e.,
C2N, C3N, molecular 2D semiconductors, carbonaceous
quantum dots and their nanohybrids with carbon nitride sheets
for solar to hydrogen generation will be scrutinized.
2. Graphene based materials as
photocatalysts
In contrast to inorganic 2D materials, carbonaceous 2D mate-
rials such as graphene, graphene oxide, and carbon nitride have
emerged as potential photocatalytic materials due to their
exceptional features including, but not limited to, high surface
area, excellent charge transport and favorable optical proper-
ties.115–117
The discussion on 2D carbonaceous structures is
incomplete without graphene, which impelled the research on
various advanced carbon based materials for numerous appli-
cations. Graphene is a monoatomically thick sp2
-hybridized 2D
carbon layer arranged in a hexagonal fashion with a continuous
p–p conjugated network. Ideally, graphene is a monoatomic
layer sheet of graphite. Graphene has emerged as an almost
magical material in the world of materials science due to its
astonishingly high surface area (2630 m2
gÀ1
), excellent electron
mobility on its surface (200 000 cm2
VÀ1
sÀ1
), high mechanical
strength (1060 GPa), high thermal conductivity (3000 W
mÀ1
KÀ1
), and favorable optical and chemical properties.118
Since the rst report on the discovery of graphene by the scotch
tape method by Geim and Novoselov,119
tremendous advance-
ments in the eld of graphitic materials have been achieved for
various applications such as batteries, solar cells, fuel cells,
biosensors, supercapacitors, catalysts, photocatalysts, etc.120–125
The isolation of a single layer defect-free graphene sheet has
been attempted by various physical means such as thermal
exfoliation, chemical vapor deposition (CVD), ultrasonication
and solvent assisted exfoliation.126
But expensive and tedious
procedures and inability to accomplish large scale production
limit commercial deployment. Chemical oxidation of graphite
with harsh oxidizing agents like KMnO4 + H2SO4, H3PO4 +
H2SO4, H2O2 + HCl, etc., partially oxidize graphite bulk material
to graphite oxide and sheets can be easily separated due to
repulsive forces between negatively charged sheets to produce
graphene oxide (GO).127,128
The GO sheets can be reduced
partially by chemical, hydrothermal or thermal treatment to
prepare reduced graphene oxide (rGO).129
Although these
methods produce graphene sheets with few defects, their
attributes are still comparable to graphene. Due to excellent
conductivity and zero band gap, which facilitate fast charge
transfer on graphene surfaces, numerous nanohybrid compos-
ites of rGO with various semiconductors such as TiO2, ZnO,
WO3, Cu2O, CuO, Fe2O3, MnO2, ZrO2, ZnS, CdS, CdSe, Bi2WO6,
MoS2, BiVO4, Sr2Ta2O7, InVO4, ZnFe2O4, etc., have been reported
for photocatalytic and other applications.60,130–133
As per the rule
of diffusion of charge (t ¼ d2
/k2
D, where d ¼ particle size, k ¼
constant, and D ¼ diffusion coefficient), two dimensional
sheets are more appealing for the water splitting process due to
the facilitation of transport of photogenerated charges to reac-
tion sites prior to recombination.134
Apart from high surface
area and a more conductive surface, it has been reported that
View Article Online

graphene can work as a macromolecular photosensitizer and
reduce the band gap of semiconductors.135
For example, ZnS is
a wide band gap semiconductor which absorbs in the UV
spectral range, while its composite with graphene (ZnS GR
nanocomposite) is fairly photoactive under visible light.136
Among various graphene/semiconductor composite materials,
cobalt oxide/graphene has proved to be an efficient catalyst.
Liang et al. demonstrated that Co3O4 nanocrystals grown on
rGO can act as a bi-functional catalyst in the oxygen reduction
reaction (ORR) and oxygen evolution reaction (OER),137
even
though both Co3O4 and rGO, taken separately, have small
affinity for these reactions. The conjugation of Co3O4 and rGO
drastically increased their electrocatalytic activity due to syner-
gistic chemical coupling effects between Co3O4 and graphene.
Nitrogen doping further enhanced the activity as well as
stability of the composite. Cobalt oxide nanoparticles and B,N-
decorated graphene (CoOx NPs/BNG) compounds displayed
better ORR and OER activity due to improved electron transfer
capacity, plenty of Co–N–C active sites and strong coupling
effects.138
The nanocomposites of rGO with hydrogen evolving
semiconductors have been found to be good hydrogen evolving
photocatalysts. The band edge position of CdS is suitable for
hydrogen evolution; however in the nanostructured form, CdS
has a tendency to aggregate, resulting in a low surface area and
therefore a lower yield. The synthesis of a nanocomposite of
CdS with GO by a hydrothermal method produces CdS nano-
particles homogeneously distributed on the surface of rGO. In
the hydrothermal treatment, residual oxygen functionalities
facilitate better attachment of nanoparticles to the sheets. It is
important to note that hydrothermal treatment does not
completely remove oxygen functionalities from the surface of
GO. The prepared rGO–CdS nanocomposites showed enhanced
water splitting characteristics.139–141
The rGO loading is crucial
for achieving an optimal hydrogen evolution rate. Excessive
loading may prevent light from reaching the semiconductor
surface and inhibit electron–hole pair generation. Furthermore,
Pt loading increases the photoefficiency of the composite. By
using a Pt loaded (1 wt%) rGO/CdS nanocomposite, a quantum
efficiency of 22.5% can be reached for hydrogen evolution at
420 nm (Fig. 4).141
N-doped graphene (N-graphene) further
increases the hydrogen evolution rate due to the creation of
a heterojunction between N-graphene sheets and CdS, which
facilitates better charge transfer. The N-graphene (2 wt%)/CdS
composite exhibited the highest H2 evolution rate which was
approximately ve times that of CdS alone.142
Core–shell structures, prepared using two or more compo-
nents, have proven themselves to be better advanced functional
materials for improved photocatalytic performance due to
synergistic attributes in terms of uniformity in size, manipulation
in core and shell composition, dispersibility, stability, and
tunability of optical, catalytic, and magnetic properties. Apart
from this, the formation of a p–n heterojunction at the inner core
and outer shell interface can be used to facilitate enhanced
charge carrier concentration on the surface.143
Many core–shell
morphologies such as rods, spheres, hollow spheres, spindles,
etc., with different arrangements of the semiconductor in the core
and shell have been prepared for solar light-driven water split-
ting.144
Graphene coated nanostructured semiconductor particles
provide promising results for the water splitting reaction due to
better dispersibility of charge over graphene sheets.145
Few layer
thick graphene sheet-wrapped particles are more efficient than
bulk nanocomposites due to transmission of light from few
layered graphene sheets to the semiconductor assembly and
better charge transportation on the surface of graphene sheets.
Few nanometer thick rGO coated anatase TiO2 nanoparticles
(graphene–TiO2 NPs) were synthesized by the surface function-
alization of TiO2 with APTMS (3-aminopropyl trimethoxysilane) to
create a positively charged TiO2 surface and subsequent wrapping
of negatively charged graphene oxide followed by hydrothermal
reduction and annealing.146
The obtained graphene–TiO2 NPs
Fig. 4 Schematics of visible light driven electron–hole pair generation in CdS and its efficient separation on reduced graphene oxide in the CdS/
rGO nanocomposite. The photogenerated electrons in the conduction band of CdS are transferred to Pt and rGO sheets, which facilitate better
charge separation and electrons are more easily accessible to protons to generate H2. Adapted with permission from ref. 141.
View Article Online

displayed excellent photoactivity under visible light for methylene
blue degradation. To exploit the advantages of the core–shell
morphology and better charge carrier mobility on rGO sheets,
Kumar et al. prepared core–shell structured CuZnO@Fe3O4
microspheres coated with rGO (rGO@CuZnO@Fe3O4) for visible
light mediated CO2 reduction.133
The TEM images of the nano-
particles showed coated rGO as an amorphous zone on crystalline
ZnO (Fig. 5). The oxide layer (mainly Fe3O4) present on the surface
of the Fe3O4 core works as a semiconductor.147
Aer the absorp-
tion of visible light, Fe2O3 can transfer photogenerated charges in
the conduction and valence bands of ZnO, while Cu acts as an
electron capturing agent to facilitate efficient transfer on rGO
sheets. The band structure of such a nanocomposite was also
suitable for water splitting and 45.5 mmol gÀ1
cat hydrogen was
evolved as a by-product.
Reduced graphene oxide can also act as an electron transport
mediator in Z-scheme photocatalytic systems. For example, a Z-
scheme heterojunction photocatalyst, consisting of BiVO4, Ru–
SrTiO3:Rh and rGO, in which BiVO4 due to its more positive
valence band facilitates water oxidation and SrTiO3:Rh due to
its more negative conduction band promotes proton reduction,
exhibited a signicantly enhanced photocatalytic water splitting
rate (Fig. 6).42
It has been found that for photoreduced graphene
oxide (PRGO), the rates of hydrogen and oxygen evolution are
much higher than those of hydrazine mediated chemically
reduced graphene oxide. This fact can be explained on the basis
of the higher hydrophobicity of PRGO, which facilitates better
Fig. 5 TEM images of (a) Fe3O4, (b) CuZnO@Fe3O4, and (c) rGO@CuZnO@Fe3O4, (d) high magnification TEM of rGO@CuZnO@Fe3O4 showing
crystalline ZnO and amorphous rGO zones, (e) HR-TEM image showing the lattice fringes and interplanar 2D spacing of ZnO and the thickness of
the rGO layer, and (f) SAED pattern of rGO@CuZnO@Fe3O4 microspheres displaying the diffraction patterns of Fe3O4 and ZnO. Reproduced with
permission from ref. 133 Copyright 2017 Elsevier.
View Article Online

transfer of electrons from the O2 evolving catalyst to the H2
reduction catalyst. Using PRGO as the electron mediator, H2
and O2 yields of 11 and 5.5 mmol hÀ1
, respectively, and an
associated turnover number (TON) of 3.2 over 24 h were
achieved.
3. Graphene oxide as a finite band
gap semiconductor for water splitting
Graphene is known as a zero band gap material. However, the
oxidation of graphene to graphene oxide (GO) transforms it into
a semiconductor. The oxidation of graphene to GO adds several
oxygen carrying functionalities, which break the conjugated p-
system of graphene due to the transformation of some sp2
carbons into sp3
carbons.148
Due to the larger electronegativity
of oxygen, the carbon skeleton of GO becomes positive and GO
behaves like a p-type material.149
The presence of conducting
sp2
and non-conductive sp3
domains on the GO sheets leads to
the creation of a band gap, which depends on the extent of
oxidation of the material. So by tuning the oxidation level, the
band gap of GO can be controlled. Yeh et al. were the rst to
demonstrate that GO can split water into hydrogen and
oxygen.75
The band gap of GO was determined to be between 2.4
and 4.3 eV, depending on the oxidation level and this band gap
was wide enough to meet the criterion of water splitting (1.23 V).
The negative conduction band afforded a hydrogen yield of 280
mmol aer 6 h of visible light irradiation, while using methanol
as a sacricial donor, the yield of hydrogen increased to 17 000
mmol aer 6 h irradiation (Fig. 7).
GO can also achieve CO2 reduction due its sufficiently
negative conduction band (À0.79 V vs. NHE) and positive
valence band (+2.91 V vs. NHE) to facilitate water oxidation for
producing the required protons.150
The decoration of GO sheets
with Cu nanoparticles further enhances the photo-efficiency
due to charge capture by Cu metal.151
Due to oxidation,
several surface defects generated on GO act as charge-
recombination sites and decrease the photocatalytic perfor-
mance of GO for water splitting. The treatment of GO with
ammonia at elevated temperature can repair these defects by
the removal of epoxy and carboxyl groups, and through the
introduction of various nitrogen functionalities, i.e., quater-
nary, pyridinic, and pyrrolic groups on the graphene sheets.
Furthermore, the band gap of GO can be reduced due to the
overlapping of p-orbitals of N and C and the delocalization of
electrons at higher energy, so the position of the HOMO gets
slightly uplied. The back oxidation of nitrogen doped gra-
phene oxide with a strong oxidizing agent (KMnO4 + H2SO4)
generates sheets having p-doped domains (due to the presence
of O on sp3
C), n-doped domains (due to the presence of N) and
interfacial channels (due to sp2
carbon). Yeh et al. showed that
N-doped graphene oxide quantum dots (NGO-QDs), due to the
presence of N and O atoms, behave like a p–n heterojunction
and the conjugated sp2
carbon skeleton acts as a charge trans-
porting tunnel (Fig. 9).76
The band gap of NGO-QDs was 2.2 eV
and linear potential scans demonstrated CBM and VBM values
Fig. 6 Two photon Z-scheme overall water splitting on the BiVO4–Ru–SrTiO3:Rh photocatalyst. BiVO4 can oxidize water to O2 by photo-
generated holes due to its positive valence band, but cannot reduce protons. The Ru–SrTiO3:Rh composite can reduce protons due to its
negative conduction band, but cannot oxidize water. The composite of BiVO4 and Ru–SrTiO3:Rh photocatalysts can sustain overall water
splitting by two photons. BiVO4, after photoexcitation, transfers electrons to the valence band of Ru–SrTiO3:Rh, which upon second excitation
reach the conduction band of Ru–SrTiO3:Rh. Reduced graphene oxide serves as a solid-state redox mediator by facilitating electron transfer
from the conduction band of BiVO4 to the valence band of Ru–SrTiO3:Rh. Adapted with permission from ref. 42.
View Article Online

of approximately À0.85 and 1.35 eV (vs. Ag/AgCl), respectively,
which validates the overall water oxidation capability. Further-
more, by changing the oxidant that governs the concentration of
oxygen on the sheets, the p-type character and band edge
position can be controlled. For example, by using HNO3 as an
oxidant for the oxidation of nitrogen-doped graphene (NG), the
CBM (À0.8 V vs. Ag/AgCl) and VBM (+1.4 V vs. Ag/AgCl) levels
were found to shi slightly (Fig. 8).152
Chen et al. demonstrated that the hydrothermal treatment of
nitrogen-doped graphene oxide quantum dots (NGO-QDs) in
ammonia transforms some pyridinic/pyrrolic groups into
amino/amide groups (Fig. 10a).153
The obtained amino/amide
groups due to the non-planar nature can donate their lone
pair of electrons to the p-conjugated network of graphene
(Fig. 10b). The intramolecular transfer of nitrogen lone pairs in
the network of sheets induces spin orbital coupling effects and
reduces the energy difference between singlet (S1) and triplet
(T1) orbitals, resulting in singlet–triplet splitting. This facili-
tates easy and efficient S1 / T1 transfer through intersystem
crossing (ISC) of excited electrons.154
The PL intensity of
Fig. 7 Band diagram of GO showing the conduction band (CB) and valence band (VB) positions relative to proton reduction and water oxidation.
The CB of GO has a higher overpotential than the level of H2 generation and photogenerated electrons are transferred to the solution phase for
proton reduction. The holes in the VB of GO are annihilated by the hole scavenger (methanol) instead of O2 generation from water oxidation due
to the higher reduction potential of methanol than water. The top-of-valence energy level was obtained using density functional theory for GO
with 12.5% O coverage, and the levels for H2 and O2 generation were obtained from the literature. Adapted with permission from ref. 75.
Fig. 8 Repairing of vacancy defects in iGO-QDs by hydrothermal ammonia treatment which introduces various nitrogen functionalities in the
iNGO-QDs. Reproduced with permission from ref. 152 Copyright 2015 Elsevier.
View Article Online

ammonia treated nitrogen-doped graphene dots (A-NGODs) was
decreased, suggesting that the removal of some oxygen func-
tionalities and slow electron–hole recombination on the N-
doped basal plane took place. Time-resolved PL decay curves
of GODs, NGODs, and A-NGODs showed average lifetimes of
excited states of 0.43, 0.54, and 1.5 ns, respectively. The longer
lifetime of excited species in A-NGODs validates the S1 4 T1
transition of electrons through intersystem crossing (Fig. 11). By
using (A-NGODs) as a catalyst, a 21% quantum yield can be
achieved by using Pt as a co-catalyst in a water/triethanolamine
mixture.
Like reduced graphene oxide, many composites of GO with
metals (Cu, Ag, Au, Pt, etc.), semiconductors (TiO2, BiVO4, CdS,
etc.), metal complexes (Ru, Co, Ir, etc.) are reported for
enhancing the hydrogen evolution rate. The surface of GO
provides the opportunity for strong interaction with metal,
metal oxide nanoparticles and other substrates which work
synergistically by facilitating charge transport, increasing
absorption and surface area to achieve overall water splitting.
The GO/TiO2 composite was found to be an effective photo-
catalyst for hydrogen production because unpaired p electrons
on GO can interact with Ti atoms on the surface of TiO2 to
generate Ti–O–C bonds, which extend light absorption in the
visible region.155–157
Due to the formation of a p–n
heterojunction in GO/TiO2 hybrids, better charge separation
can be achieved which facilitates efficient water splitting
(Fig. 12).
4. Sensitization of graphene-based
materials
Due to the presence of an ample number of functional groups,
GO provides an opportunity for covalent functionalization of
metal complexes on its surface. Metal phthalocyanine and
ruthenium complexes have been successfully anchored onto
GO; however due to the presence of mainly epoxide function-
alities on basal planes, lower loading was achieved. Therefore,
efforts are underway for the covalent functionalization of basal
planes by breaking epoxide bonds followed by the attachment
of metal complexes. Kumar et al. demonstrated the attachment
of phthalocyanines (Co,158
Cu159
and Fe160
) and polypyridyl metal
complexes (Ru161,162
) on GO sheets with high loading by func-
tionalizing epoxide moieties with chloroacetic acid and subse-
quent covalent attachment of homogeneous metal complexes
for various applications, i.e., CO2 photoreduction, thiol oxida-
tion, CO2 to DMF formation, etc. It has been found in all cases
that the photocatalytic performance of metal complex-loaded
GO was increased manifold due to better charge injection in
the conduction band of GO. Hexanuclear Mo metallic clusters
([Mo6Br14]2À
),163
which are very photoactive and able to transfer
multiple electrons in photocatalysis processes, were also
immobilized on GO by taking advantage of the labile nature of
their apical halogen atoms, which can be easily replaced by
oxygen present on graphene sheets (Fig. 13).164
The sensitization of undoped and doped graphene with
metal complexes or organic dyes further improves the visible
light absorption as well as photocatalytic performance.165
A
copper complex immobilized on N-doped graphene (GrN700–
CuC) was found to be a good photocatalyst for CO2 reduction to
methanol (1600 mmol gcat
À1
) in water/DMF/triethylamine solu-
tion under visible light irradiation.166
Wang et al. reported
covalently immobilized manganese phthalocyanine to reduced
graphene oxide by using a glycinato linker (Fig. 14).167
The
synthesized photocatalyst was used for the production of
hydrogen from water in the presence of platinum. The yield of
hydrogen was found to be 8.59 mmol mgÀ1
during this study.
Metal-free dyes due to their inexpensive nature, degradability
and wide absorption range have proven to be good photo-
sensitizers for the hydrogen evolution reaction. Eosin Y, an
organic acidic dye, binds with reduced graphene oxide (EY–
RGO) through hydrogen bonding and p–p interaction to
produce a photocatalytic material for hydrogen generation in
the presence of Pt nanoparticles decorated on the surface of
reduced graphene oxide (Fig. 15).168,169
The presence of both Pt
and RGO was crucial for the production of hydrogen and by
using a EY–RGO/Pt photocatalyst the rate of hydrogen produc-
tion reached was about 10.17 mmol hÀ1
, corresponding to an
apparent quantum yield (AQY) of 4.15% at 420 nm. The superior
photocatalytic activity was assumed to occur due to efficient
charge separation on the surface of RGO and subsequent
Fig. 9 Schematics of p–n heterojunction formation on NGO-QDs
during hydrothermal treatment of GO-QDs with NH3. Reproduced
with permission from ref. 76 Copyright 2014 Wiley-VCH.
View Article Online

capture by Pt nanoparticles. Furthermore, to maximally harvest
the solar spectrum, EY and Rose Bengal (RB) were graed on the
surface of RGO (ER-G/Pt) and the rate of hydrogen evolution
increased signicantly. By using ER-G/Pt, the hydrogen
production rate was increased up to 36.7 mmol hÀ1
,
corresponding to a quantum yield (QY) of 9.4% at a 420 nm
wavelength.79
5. Doped graphene photocatalysts
Although graphene is a zero band gap semimetal in which
quasiparticles obey a linear dispersion relationship between the
energy and momentum, i.e., the conduction band touches the
Fig. 10 Schematic of (a) hydrothermal cleavage of pyridinic and pyrrolic rings in NGODs to create epoxy, carbonyl, and carboxylic groups for
decorating the basal plane and periphery of the A-NGODs with amino and amide groups. (b) An orbital diagram showing the overlaps of the
nitrogen lone pair orbital in nitrogen functionalities with the aromatic p-system. The more effective overlap of the lone pair orbital of amino
(–NH2) groups in A-NGODs with the aromatic p-system of sheets than the planar pyridinic-N atom lone pair orbital has been revealed.
Reproduced with permission from ref. 153 Copyright 2016 Wiley-VCH.
Fig. 11 The band structure diagram of NGODs and A-NGODs shown
relative to the redox potential of water for H2 and O2 generation. The
small energy difference between singlet (S1)–triplet (T1) splitting of the
CBM of A-NGODs facilitates the S1 4 T1 transition of electrons via ISC,
which prolongs the lifetime of the excited electrons for H2 production
from water reduction. Reproduced with permission from ref. 153
Copyright 2016 Wiley-VCH.
Fig. 12 The band gap narrowing of TiO2 due to Ti–O–C bonding
between unpaired p-electrons on GO and surface Ti atoms of TiO2,
which facilitates the light absorption profile of TiO2. Reproduced with
View Article Online

valence band at a single Dirac point, there are various routes to
create a band gap in graphene (Fig. 16a). By lowering the size of
graphene sheets, i.e., nanoribbons and nanodots, a band gap is
opened in graphene due to the quantum connement eﬀect.
Furthermore, the number of sheets is also an important
parameter which inuences the electronic environment of
another sheet. The charge density on the edge of sheets can be
increased in comparison to the basal plane. The presence of
Fig. 13 A schematic illustration of Mo6 clusters decorated on GO nanosheets; the right side image depicts the structure of the Mo6 cluster exhibiting
the positions of inner and apical ligands. High resolution XPS spectra in the Br3d region for Cs2Mo6Br8
i
Br6
a
(ai) and GO-Cs2Mo6Br8
i
Brx
a
(aii) composites,
and (TBA)2Mo6Br8
i
Br6
a
(bi) and GO-(TBA)2Mo6Br8
i
Brx
a
(bii) composites showing a decrease in the signal intensity of apical bromine atoms, suggesting
their attachment to GO by the replacement of bromine atoms with oxygen on GO. Reproduced with permission from ref. 164 Copyright 2015 Elsevier.
Fig. 14 Outline of the electron transfer from MnPc to a graphene
sheet for hydrogen evolution. Reproduced with permission from ref.
167 Copyright Royal Society of Chemistry.
Fig. 15 Bonding of Eosine dye (EY) on rGO sheets through p–p
interaction. EY transfers visible light photogenerated electrons to rGO
sheets, which due to higher mobility increases charge separation.
Electrons on rGO sheets are captured by Pt and used for hydrogen
generation from water. Adapted with permission from ref. 168.
View Article Online

defects like vacancies and dislocations also opens the band gap
by inducing additional electronic states.170
The introduction of
various elements like H, O, N, F, etc., also affects the electronic
properties of sheets. For example, uorographene, a 2D sheet of
graphite uoride, is an insulator and its band gap can be tuned
by controlling the number of uorine atoms (Fig. 16b).171
The
attachment of uorine atoms to sp2
hybridized carbon distorts
the hybridization of carbon to sp3
and shis the Dirac point.
Theoretical calculation revealed that the band gap of partially
uorinated graphene (from C32F to C4F) can be tuned from 0.8
to 2.9 eV.172
However, experimental results showed an optical
gap >3.8 eV (Fig. 17).173
Partially chlorinated and brominated
graphene counterparts have also been reported.174
The uo-
rographene provides an opportunity for further functionaliza-
tion of graphene with high selectivity. Cyano groups were
introduced in a scaffold of the graphene skeleton by the
replacement of uorine atoms and the resulting cyanographene
possesses high negative charge as revealed by zeta potential
measurements.175
The replacement of F atoms in uo-
rographene by hydroxyl groups transforms graphene into
a room temperature organic magnet and the magnetic proper-
ties can be tailored depending on the F/OH ratio.176
In addition,
several covalent functionalization approaches have been
utilized to form various functional groups such as –SH, –OH,
alkyl, etc., on the surface of graphene.177–179
Tailoring the structure of pristine graphene by substitution
with heteroatoms like B, N, P, F, etc., inuences the charge
distribution on the sheets and represents an effective way to
modify the electronic, surface, optical and electrochemical
properties. The dopant atoms can either work as donors or
acceptors, depending on their electronegativity with respect to
the carbon atom. Various heteroatoms like N, B, and P have
been introduced into the graphene skeleton by using different
precursors. Among them N doping has been found to show the
best results due to the higher energy level of the N1s orbital,
which uplis the energy level of the newly hybridized orbital. N-
doping contributes mainly three kinds of nitrogen to the
Fig. 16 Structure of (a) graphene sheets showing the planar sheets of
sp2
hybridized carbon atoms arranged in a hexagonal fashion, and (b)
fluorographene sheets of sp3
carbons atoms arranged in a chair
conformation. Adapted from ref. 171.
Fig. 17 The opening of the band gap in fluorographene. (a) NEXAFS spectra of pristine graphene and fluorographene with two different contents
of fluorine. The dashed lines at 284.1 and 287.9 eV mark the leading edges of the p* resonance for the pristine and fluorinated samples,
respectively. (b) The PL emission of the pristine graphene and fluorographene dispersed in acetone using 290 nm (4.275 eV) excitation at room
temperature. The optical images in the inset show the blue PL emission, which persists for approx. 30 s after the excitation laser is turned off.
(Reproduced with permission from ref. 173 Copyright 2011 American Chemical Society).
View Article Online

graphenic structure: (1) pyridinic nitrogen, (2) quaternary
nitrogen, and (3) pyrrolic nitrogen (Fig. 18).180
Pyridinic nitro-
gens due to their position and availability of their lone pairs in
the p orbital can participate in conjugation and increase the
charge density of sheets; the Fermi level shis above the Dirac
point, distorts the symmetry of the graphene sub-lattice and
creates a band gap in the N-doped sheets.181
The value of the
band gap is strongly dependent on the amount of nitrogen; by
manipulating the concentration of nitrogen atoms, the band
gap can be tuned.
N-doped graphene can be synthesized by various methods
such as CVD, arc discharge method, and the thermal treatment
of GO or rGO with nitrogen rich sources like ammonia, amines,
hydrazine hydrate, urea, melamine, nitrogen, etc.184
The most
commonly used nitrogen source is ammonia even though the
nitrogen content with ammonia treatment remains low
(Fig. 19). Numerous composite materials such as Co/Co9S8 (ref.
185) with doped-graphene have been utilized for photocatalytic
applications, i.e., dye degradation, water splitting, solar cells,
etc. Like N-doping, P-doping also induces band gap opening in
graphene and can afford water splitting. In a recent report, 1.3
at% P-doped graphene was synthesized by thermal annealing of
graphene with phosphoric acid and showed improved capaci-
tance and cycling stability.186
Biological carbon precursors
modied with phosphoric acid can also generate P-doped gra-
phene. For example, phosphoric acid modied alginate, aer
thermal annealing, gives P-doped graphene having high P
content (17.16 at% for (P)G-1).187
The band gap of P-doped
graphene was estimated to be 2.85 eV and afforded 282 mmol
gÀ1
hÀ1
of H2. Some doped-graphene sheets demonstrate
enhanced electrocatalytic performance comparable to that of
precious Pt metal.188,189
The doping of graphene with two dopant
atoms (with reverse electronegativity like B and N) can accu-
mulate synergistic effects to boost electrocatalytic performance
in the oxygen reduction reaction (ORR).190–192
However, most of
the reports focused on the use of doped graphene for the ORR
Fig. 18 Structure of N-doped graphene showing various types of N
atoms according to bonding configuration. Adapted with permission
from ref. 182.
Fig. 19 Schematic diagram for the preparation of N-doped graphene with different N states. N-RG-O 550, 850, and 1000
C were prepared by
annealing of GO powder at temperatures of 550
C, 850
C, and 1000
C in the presence of NH3. PANi/RG-O and Ppy/RG-O were prepared by
annealing of PANi/G-O and Ppy/G-O composites at 850
C. Reproduced with permission from ref. 183 Copyright 2012 Royal Society of
Chemistry.
View Article Online

and only a few reports are available for the hydrogen evolution
reaction (HER). In this avenue, theoretical calculations showed
that N- and P-co-doped graphene should be the best choice for
the HER due to down shiing of the valence band, which may
result in the highest HER activity over various singly doped and
co-doped graphene materials. To achieve this goal, N- and P-co-
doped graphene (N,P-graphene-1) was prepared by annealing of
melamine (nitrogen source), triphenylphosphine (P source) and
GO (graphene oxide sheets) in a 10 : 10 : 1 mass ratio under
argon at 950
C for 3 h. The N,P-doped graphene-1 displayed
a 10 mA cmÀ2
HER current density at an overpotential of
$420 mV, which was much lower than that of single atom
doped N- and P-graphenes.193
For the doping process, ionic
liquids can also be used as a source of heteroatoms and they
also provide porosity to materials. For example, thermal
annealing of polyaniline-coated GO and ammonium hexa-
uorophosphate (NH4PF6) produced nitrogen, phosphorus, and
uorine tri-doped graphene with a porous structure.194
In this
process, NH4PF6 serves as a source of heteroatoms as well as
a so template and its degradation produced gases, which
provided porosity to the material. The metal free N-, P-, and F-
tri-doped graphene multifunctional catalyst furnished
hydrogen and oxygen gas with a production rate of 0.496 and
0.254 ml sÀ1
, respectively.
The OER is the main constraint in the water splitting reac-
tion, which requires noble metal oxides such as IrO2 or RuO2 to
overcome the sluggish kinetics of the OER. As an alternative,
NiCo2O4 is a suitable choice over these noble metals due to its
low cost and wide abundance. The performance of NiCo2O4 can
be further improved by hybridization with graphenic materials.
In this avenue, Chen et al. reported on the synthesis of a 3D N-
doped graphene NiCo2O4 nanocomposite by using in-plane
porous graphene and subsequent doping and controlled
growth of NiCo2O4 for an enhanced OER (Fig. 20).195
The pyrolysis of carbonaceous materials at an elevated
temperature under an inert atmosphere gives defect-rich gra-
phenic carbon and use of an additional nitrogen source or the
precursor itself can produce N-doped carbon. Due to the pres-
ence of a band gap, high surface area, better light absorption
and easy synthesis from earth abundant materials, doped
carbon represents an interesting alternative to replace graphene
materials in photocatalytic applications. By choosing appro-
priate precursors, the amount of the dopant can be controlled
in the samples. For the synthesis of doped carbon materials,
various carbon sources rich in heteroatoms can be utilized.
Biomass rich in nitrogen content also provides opportunities
for the synthesis of derived doped carbon from cheap and
environmentally benign sources.196
Similar to heteroatom co-
doped graphene, the co-doping of carbon with diﬀerent
heteroatoms creates variable band gaps and numerous active
sites on the surface, which lead to the enhancement of catalytic
and photocatalytic performance.197
For example, the carbon-
ization of melamine (N source), glucose (C source) and sul-
phuric acid (S source) produced N-, S-doped carbon exhibiting
a high surface area.198
Due to heteroatom doping, several sites
are evolved on the surface of doped carbon and serve as active
sites for various catalytic reactions. For example, N-, P-, and S-
doped hollow carbon shells with a very high surface area were
synthesized by thermal treatment of poly-
(cyclotriphosphazene-co-4,40
-sulfonyldiphenol) (C, N, P, and S-
doping source) and ZIF-67 (source of the carbon frame-
work).199
In recent years, metal organic frameworks (MOFs) have
become popular for the synthesis of heteroatoms, particularly
N-rich carbon.200,201
The advantages of carbon derived from
MOFs are high N content, high porosity, periodicity of defects
and possibility of manipulation of the band gap, which make
them excellent metal free photocatalytic and electrode materials
for hydrogen and oxygen evolution. Zeolitic imidazolate
framework (ZIF) based MOFs are widely utilized for the
synthesis of N-rich carbon materials. Zhao et al. reported the
synthesis of nitrogen-rich carbon (ZNG) by the carbonization of
ZIF-8 at 800–1000
C.202
The hydrogen evolution rate from
a sample prepared at 1000
C carbonization temperature was
found to be 18.5 mmol hÀ1
under visible light in the presence of
triethanolamine and 3 wt% Pt co-catalysts. This catalyst showed
excellent photocatalytic performance even in the absence of Pt
with a hydrogen evolution rate of $5.8 mmol hÀ1
. CoP nano-
particle embedded N-doped hollow carbon nanotubes/
polyhedra derived from a core–shell structured ZIF-8@ZIF-67
MOF displayed increased water splitting eﬃciency.203
Several
nanocomposite materials of semiconductors and doped carbon
have been employed for various applications particularly elec-
trochemical and photochemical hydrogen evolution.204–207
For
example, CoO nanoparticles embedded in a N-doped carbon
layer were synthesized by thermal pyrolysis of polypyrrole-
coated Co3O4 nanoparticles at 900
C; this material exhibited
excellent activity for the ORR and OER which was assumed to be
Fig. 20 Fabrication of N-doped graphene–NiCo2O4 3D hybrid cata-
lyst (Notations PG, PNG and PNG-NiCo signify porous graphene,
porous N-doped graphene, and porous N-doped graphene–NiCo2O4
hybrid, respectively). Reproduced with permission from ref. 195
Copyright 2013 American Chemical Society.
View Article Online

due to the porous nature of doped carbon and synergistic
interaction.208
The main challenge in electrocatalytic water
splitting is the replacement of the expensive noble metal plat-
inum. Metal carbide (particularly Mo2C) seemed to be an
interesting alternative to Pt because of similar d-band electronic
density, high electrical conductivity, and good hydrogen–
adsorption properties. In this respect, ultra-small molybdenum
carbide (Mo2C) nanoparticles embedded in nitrogen-rich
carbon Mo2C@NC displayed remarkable catalytic activity and
stability over a wide range of pH (pH 0–14) for the electro-
catalytic HER.209
The Mo2C@NC produced hydrogen at a rate of
34.5 mmol hÀ1
, corresponding to 100% faradaic efficiency close
to the theoretically achievable value. The dramatically high
performance of this catalyst was believed to be due to the strong
electron-withdrawing features of N dopants, which facilitated
electron transfer in the direction Mo2C / C / N, so that
neighboring C atoms could act as donors and acceptors to
provide active sites. Another report also demonstrated that
Mo2C embedded in N-doped carbon achieved better electro-
catalytic water splitting performance.210
The controlled
synthesis of N-doped porous carbon sandwiched reduced gra-
phene oxide (N-PC@G) was achieved by the pyrolysis of ZIF-8
sandwiched GO (Fig. 21). The structure obtained by maintain-
ing a GO concentration of 0.02 g in reaction and thermal
treatment at 900
C afforded the highest surface area (1094.3 m2
gÀ1
) and the obtained catalyst showed better OER and ORR
activity due to better charge transport of holes generated in
doped carbon on rGO sheets.211
6. Carbon nitride based
photocatalysts for hydrogen evolution
Polymeric semiconductor materials such as polyaniline and
polyparaphenylene212
have been used for hydrogen production
even though they absorb UV light and the associated quantum
efficiency is too low. In recent years, carbon nitride, the oldest
reported synthetic organic polymer, has re-emerged as a mate-
rial of choice due to its specic chemical and physical proper-
ties such as a low band gap and a suitable band edge position
for efficient water splitting, nontoxic nature, wide stability, easy
Fig. 21 Schematic illustration of the synthesis steps of sandwich-like structured N-doped porous carbon@graphene (N-PC@G). Reproduced
with permission from ref. 211 Copyright 2016 Elsevier.
Fig. 22 Structure of graphitic carbon nitride based on (a) s-heptazine (gh-C3N4) and (b) s-triazine units (gt-C3N4). The s-heptazine network gh-
C3N4 is about 30 kJ molÀ1
more stable than gt-C3N4. Adapted with permission from ref. 213 and 214.
View Article Online

synthesis and processability.87
In the world of 2D materials,
carbon nitride competes with graphene because of a similar
graphitic structure, high surface area, astonishing optical and
chemical properties and ease of synthesis. Carbon nitride
(C3N4) is a 2D polymeric semiconductor consisting of inter-
connected tri-s-triazine units via tertiary amines. Carbon nitride
exists in several allotropic forms (C3N4, that is, a-C3N4, b-C3N4,
cubic C3N4, pseudocubic C3N4, g-h-triazine, g-o-triazine, and g-
h-heptazine) depending on the basic monomer unit and the
pattern of linkage, which governs the morphology, crystallinity
and even photoactivity. Graphitic carbon nitride possessing
a layered graphite-type structure is the most stable form.
Graphitic carbon nitride (g-C3N4) is either composed of triazine
(C3N3) units in which case it is amorphous and less stable or tri-
s-triazine (heptazine, C6N7) units, which results in a crystalline
and more stable compound (Fig. 22). The synthesis of carbon
nitride from its precursor always aﬀords carbon nitride with
a heptazine structure due to its higher stability. The continuous
N-bridged tris-s-triazine ring structure gives rise to graphite type
p-conjugated planar sheets. Due to the high degree of poly-
merization of tri-s-triazine units, g-C3N4 exhibits astonishing
thermal stability (up to 600
C) and chemical stability (stable in
acidic and basic media). Triazine units containing amorphous
carbon nitride (gt-C3N4) can be synthesized by the reaction of
cyanuric chloride and lithium nitride.213,214
Tri-s-triazine units
containing carbon nitride frameworks can be isolated by the
polycondensation of nitrogen rich precursors such as urea,
cyanamide, dicyandiamide, thiourea, formamide, melamine,
CCl4/ethylene diamine, guanidine, etc. Due to alternate repeti-
tion of carbon and nitrogen in carbon nitride, it has a medium
band gap of 2.7 eV and absorbs in the blue region (420–450 nm)
of visible light. The position of the conduction band at À1.1 V
(vs. NHE at pH 0) and valence band at +1.6 V (vs. NHE at pH 0)
makes it a suitable candidate for water splitting (Fig. 23).51,215,216
Large scale production using urea provides an opportunity
for the realization of a catalyst for future use; however mass loss
is more important than with other precursors although it can be
limited by condensation in closed systems or co-condensation in
a cage of porous material such as silica.217,218
The synthesis of
carbon nitride materials from urea with a high g-C3N4 yield and
variable band gap can also be achieved using eutectic molten
alkyl halide salts.219
The annealing temperature also controls the
photoactivity of g-C3N4. Generally, most of the reports described
the synthesis of carbon nitride at 550
C. However, a report by
Papailias et al. showed that a sample annealed at 450
C was
more photoactive than samples annealed at 550
C for NO
degradation.220
It is proposed that polycondensation at
a controlled temperature produces nitrogen vacancy defects,
which prevent the intrinsic radiative decay of charges and
populate the conduction and valence bands with more available
electrons and holes, respectively.221
A TiO2/C3N4 composite,
prepared at high calcination temperature (600
C), was found to
be optimal due to the creation of oxygen vacancies and Ti3+
defects, which promote Z scheme transition instead of direct
transition.222
Besides the annealing temperature, the annealing
environment also inuenced the photocatalytic performance of
g-C3N4. g-C3N4 containing carbon vacancies and nanoholes,
prepared by two step heating (i) at 650
C under argon and (ii) at
550
C in air, displayed a high H2 evolution rate (5261 mmol hÀ1
gÀ1
) with a quantum eﬃciency of 29.2% at 400 nm.223
Fig. 23 Band structure diagram of g-C3N4 showing the positions of conduction and valence bands and relative oxidation and reduction
potentials of water.
View Article Online

Interestingly, the deliberate introduction of defects into carbon
nitride sheets improved the photocatalytic performance. Niu
et al. reported that rapid heating (5 min) of carbon nitride at
650–700
C followed by fast cooling promoted defects in the CN
structure and the band gap was reduced to 2.17 eV which is low
enough to absorb a large part of the visible spectrum.224
Furthermore, due to the band gap and presence of nitrogen-
rich sites on carbon nitride, it can promote various oxidation
and reduction reactions such as alkylation, oxidative coupling,
hydrogenation (phenol to cyclohexanone), NO decomposition,
oxygen reduction reaction (ORR), and CO2 activation under
metal-free conditions.225–230
The presence of various kinds of
nitrogen (especially pyridinic) in g-C3N4 makes it more electron-
rich and provides active sites to facilitate base promoted
nucleophilic reactions like the oxidation of alcohols to alde-
hydes by activating O2,231
transesterication of keto-esters,232
and three component coupling (benzaldehyde, piperidine, and
phenylacetylene) to propargylamine.233
It is important to note
here that the three component coupling reaction proceeds only
in the presence of strong bases such as butyl-lithium, organo-
magnesium reagents, or lithium diisopropylamide. Despite
being a semiconductor, carbon nitride also suffers from the
drawback of charge recombination and limited light absorption
in the lower wavelength region, which restricts its photoactivity.
Furthermore, due to a more reductive conduction band than the
water reduction potential (0.00 V vs. NHE at pH 0), g-C3N4
showed excellent hydrogen evolution activity; in contrast, the
valence band is relatively weakly oxidative, compared to the
water oxidation potential (+1.23 eV), leading to low activity for
oxygen evolution activity. Many efforts have been devoted to
improving the performance of carbon nitride by adding various
metals (Pd, Cu, Ru, Fe, Pt, and Ni), and inorganic semi-
conductors (InVO4, Fe3O4, Cu2O, TiO2, NaNbO3, Mn0.8Cd0.2S,
Bi2WO6, ZnO, WO3, ZnCr2O4, CoP, Co2P, CuCo2O4, Ni2P, WS2,
BiPO4, MoS2, In2O3, CeO2, ZnS, etc.), graphene oxide, graphene,
fullerene, carbon dots, organic polymers, metal complexes,
metal free dyes, etc.92,215,234–243
The composite materials of
carbon nitride with semiconductors not only increase the
photocatalytic efficiency by synergistic mechanisms, but also
improve the photostability. For example, AgPO4, an excellent
oxygen evolving catalyst, suffers from the drawback of self-
reduction to silver.244
Furthermore, a positive conduction
band edge comparable to the proton reduction potential and
a tendency to make large particles make it inappropriate for
hydrogen evolution. However, upon hybridization with g-C3N4,
it becomes quite stable with better photoactivity due to the
formation of heterojunctions and smaller particles. Yang et al.
prepared a AgPO4 and carbon nitride hybrid (AgPO4/ECN400) by
a two-step method using cyanuric acid and melamine as
precursors of g-C3N4 and the subsequent growth of AgPO4 by
reaction of a Ag+
ion source and NaPO4.245
The synthesized
AgPO4/ECN system was found to be very stable and exhibited
excellent photocatalytic performance for rhodamine B (RhB)
degradation and O2 evolution. Previously proposed mechanisms
Fig. 24 ESR spectra of radical adducts trapped by DMPO in an (a) aqueous dispersion of Ag3PO4, (b) aqueous dispersion of ECN400, (c) Ag3PO4
and ECN400 in the presence or absence of AgNO3, and (d) ECN, Ag3PO4, and ECN400 methanol dispersions. Reprinted with permission from ref.
245 Copyright 2015 Wiley-VCH.
View Article Online

suggesting a transfer of electrons from the CB of g-C3N4 to the CB
of AgPO4 and holes from the VB of AgPO4 to the VB of g-C3N4 were
not convincing, due unfavorable charge transfer and inability of
accumulated holes in the VB of g-C3N4 to facilitate dye oxidation
and O2 evolution.246,247
The negligible O2 yield and RhB degra-
dation activity using pristine g-C3N4 are due to its inability to
facilitate the water oxidation process. Dye degradation experi-
ments by introducing benzoquinone (BZQ) (an O2cÀ
radical
scavenger), tert-butyl alcohol (OHc scavenger) and Na2-EDTA
(direct hole scavenger) revealed that O2cÀ
was the dominant
active species responsible for dye degradation. The investigation
of the reaction mechanism with the help of ESR using 5,5-
dimethyl-1-pyrroline-N-oxide (DMPO) as a trap agent showed the
signals of DMPO–OHc and DMPO–O2cÀ
adducts, which in turn
indicate the generation of OHc and O2cÀ
radicals under visible
light irradiation with increasing signal intensity during the
duration of irradiation. Interestingly, for AgPO4 and ECN400 in
the presence of AgNO3, the signals from DMPO–OHc disappeared
aer irradiation due to the generation of OHc radicals and the
intensity of the signal increased during the irradiation time,
suggesting that fewer OHc radicals were generated at an earlier
stage; the absence of an ESR signal in the dark demonstrated that
the presence of light is crucial. Additionally, the signals of
DMPO–O2cÀ
adducts were observed in pristine AgPO4, ECN400
and the AgPO4/ECN400 hybrid conrming the generation of
superoxide radicals (Fig. 24). On the basis of these results, a Z
scheme mechanism was proposed for higher photocatalytic
performance. Initially, under visible light irradiation, electrons
generated in the CB of carbon nitride are transferred to the CB of
AgPO4. These electrons reduce Ag+
to Ag and the system turns
into a three phase Ag3PO4/Ag/ECN system. Subsequently, elec-
trons generated in the conduction band of AgPO4 and holes
generated in the valence band of carbon nitride recombined at
the Ag interface, resulting in electrons in the conduction band of
g-C3N4 (ECN400) and holes in the valence band of AgPO4 (Fig. 25).
The electrons in the CB of ECN400 can reduce O2 to an O2cÀ
radical which is transformed in to a cOH radical aer the
abstraction of a proton from water, while holes in the VB of AgPO4
were utilized for the direct oxidation of RhB and water. The
proposed Z scheme was further supported in follow up studies.248
Fig. 25 Schematic for the energy-level diagram of (a) Ag3PO4 and (b) ECN, (c) conventional mechanism, and (d) proposed Z-scheme mechanism
for charge transfer. The band-gap values of Ag3PO4 and ECN were estimated to be 2.45 and 2.70 eV, respectively. The EVB of Ag3PO4 and ECN
was calculated to be 2.90 and 1.53 eV vs. NHE, respectively, whereas their corresponding ECB was 0.45 and À1.17 eV vs. NHE, respectively.
Reproduced with permission from ref. 245 Copyright 2015 Wiley-VCH.
View Article Online

7. Doped-carbon nitride for
photocatalysis
The doping of carbon nitride with heteroatoms is appealing
because heteroatoms can inuence charge density on sheets,
which affects visible light absorption properties. Among various
possible doping agents, ionic liquids having a heteroatom
counter ion are considered the best candidates for the doping of
carbon nitride because of high boiling points, negligible vapor
pressure and high stability (up to 400
C) so that they can
participate in the C–N condensation step during polymeriza-
tion. Phosphorous doping is appealing because the replace-
ment of C in C3N4 sheets with electron rich P atoms will lead to
the addition of a lone pair of electrons in the sheet's p conju-
gated system (Fig. 26). Zhang et al. synthesized phosphorus-
doped carbon nitride by heating dicyandiamide and BmimPF6
(1-butyl-3-methylimidazolium hexauorophosphate) precur-
sors.249 31
P NMR analysis of the developed materials indicated
that some of the carbon atoms were replaced by P atoms.
Additionally, XPS analysis revealed 3.8 at% of phosphorus in P-
doped carbon nitride. Due to P doping, the band gap of P-doped
C3N4 was reduced down to 1.60 eV, so that it can absorb almost
in the whole visible spectrum. The band gap value of materials
strongly depends on the amount of P doping and can be tuned
by controlling the precursor concentration. It is important to
note here that the precursor of the doping material should be
stable and participate in the polycondensation step. For
example, P-doped carbon nitride can also be synthesized using
relatively inexpensive hexachlorotriphosphazene (NPCl2)3 as
a source of phosphorus and guanidinium hydrochloride
(GndCl) as a precursor for the C3N4 framework.250
The synthe-
sized catalyst showed better performance for visible light
promoted hydrogen evolution in the presence of a Pt co-catalyst
and triethanolamine as a sacricial donor. The P-doped C3N4
synthesized using 10 wt% hexachlorotriphosphazene at 550
C
(P10-550) afforded the highest rate of hydrogen evolution (50.6
mmol hÀ1
), which was 2.9 fold higher than that of pure g-C3N4.
Similarly, a P-doped nanotubular structure was prepared using
melamine and sodium pyrophosphate (phosphorous atom
source) followed by heating at 500
C and then used for
hydrogen production from water splitting.251
The activity of the
doped catalyst can be further improved by adding metal nano-
particles like nickel nanoparticles. Indeed, it has been demon-
strated that P-doped carbon nitride decorated with Ni
nanoparticles (Ni@g-PC3N4) displayed better photoactivity than
P-doped C3N4 for visible light driven reduction of nitro
compounds.252
P-doped C3N4 was synthesized by using
BmimPF6 as a phosphorus source and dicyandiamide as a hep-
tazine framework source;249
Ni@g-PC3N4 was prepared by
introducing specic wt% of Ni nanoparticles into dicyandia-
mide and BmimPF6 precursors followed by heating at 550
C.
5 wt% loading of Ni nanoparticles was found to be optimum to
achieve the complete transformation of nitroaromatics to
anilines within 8 h of visible light irradiation.
Similar to P-doping, boron and uorine co-doped meso-
porous carbon nitride (CNBF) was synthesized by thermal
polymerization of dicyandiamide and 1-butyl-3-
methylimidazolium tetrauoroborate (BmimBF4) in which
BmimBF4 itself serves as a source of B and F but also as a so
template for improving the surface area.253
The CNBF catalyst
promotes the selective transformation of cyclohexane to cyclo-
hexanone under thermal and oxidative conditions. Although
boron doping of carbon nitride gave good results for organic
transformation, these structures were found to be inactive for
water splitting. This has been assigned to the detrimental effect
Fig. 26 Chemical structure of P-doped C3N4 showing Lewis acid and Lewis base sites created by P doping and –NH2 groups present on the
periphery. The pyridinic nitrogen due to the presence of lone pairs also provides basic sites. Reproduced with permission from ref. 250 Copyright
2015 Royal Society of Chemistry. The structure on the right side depicts the electron transfer from a P atom to the p-conjugated network
responsible of the band gap reduction of P-doped C3N4.
View Article Online

of dopant debris, which behave as deep trap-sites and can also
unfavorably distort the electronic structure. However, the care-
ful selection of dopant precursors can overcome this problem.
For instance, B-modied g-C3N4 synthesized by thermal
annealing of low cost urea and Ph4BNa exhibited good
hydrogen activity for the hydrogen evolution reaction.254
Boron
acts as a Lewis acid, while nitrogen atoms on sheets work as
intrinsic Lewis base sites which generate frustrated Lewis pairs
and favor the fast splitting of light-generated excitons.255
So,
these Lewis pairs on sheets facilitate the formation of hydrogen
adatom centers on the acid–base sites, which promote the
surface activity of the catalyst.256
Apart from BmimBF4, ammo-
niotrihydroborate (BNH6) and boric acid B(OH)3 have also been
used as boron sources for the synthesis of B-doped carbon
nitride.257,258
Nitrogen-doped carbon nitride was also prepared
by using EMIM-dca (1-ethyl-3-methylimidazolium dicyanamide)
and 3-MBP-dca (3-methyl-1-butylpyridine) ionic liquids.259
The
carbon to nitrogen ratio of g-C3N4 is 3 : 4 and the main chal-
lenge is to increase the nitrogen content, which may potentially
reduce the band gap. To achieve this goal, Vinu et al. prepared
Fig. 27 (A) Carbon K-edge and nitrogen K-edge NEXAFS spectra of (a) MCN-8 and (b) g-C3N4 (inset: proposed molecular structure). (B) High-
resolution XPS of MCN-8 in C1s and N1s regions. Reproduced with permission from ref. 260 Copyright 2017 Wiley-VCH.
Fig. 28 Synthesis outline of F-doped carbon nitride (CNF) showing out of plane F atoms attached to corner C atoms. Reproduced and redrawn
with permission from ref. 262 Copyright 2010 American Chemical Society.
View Article Online

new nitrogen-rich carbon nitride (MCN-8) having a C3N5 stoi-
chiometry by using 3-amino-1,2,4-triazole as a source of
nitrogen (Fig. 27). Due to N-doping, the band gap of carbon
nitride was reduced to 2.2 eV and the surface area was increased
up to 296.7 m2
gÀ1
.260
The obtained catalyst was able to split
water into hydrogen at a signicant rate (801 mmol of H2 under
visible-light irradiation for 3 h) without the use of any external
dopant. Furthermore, carbon nitride polymer with a high N
stoichiometric ratio (C3N6) containing diaminotetrazine units
has been reported by the same group for improved water
splitting.261
Instead of the addition of expensive dopants, some cheap
nitrogen and heteroatom precursors can provide doping
elements in the polycondensation step. For example, uorine-
doped carbon nitride (CNF) can be synthesized by using NH4F
and dicyandiamide precursors (Fig. 28).262
XPS and 19
F solid
state NMR analysis conrmed the doping of uorine ($3 at%)
and theoretical calculation showed that the uorine atom was
attached to a carbon atom in an out of plane fashion. The
attachment of F atoms to corner and bay carbon atoms shis
the LUMO and HOMO positions and modies the redox
properties in such a way as to promote heterogeneous photo-
catalysis (Fig. 29). The band gap value of CNF was determined
to be 2.63 eV and the hydrogen evolution rate of the CNF
catalyst using Pt as a co-catalyst from a water/triethanolamine
mixture under visible light (l 420 nm) was 2.7 times higher
than that of unmodied g-C3N4. A recent study demonstrated
that the hydrothermal treatment of g-C3N4 with dilute HF does
not lead to F doping. However, structural distortion was
induced in the CN framework, which creates delocalized elec-
tronic states that aﬀect the charge distribution pattern on CN
sheets. Due to this distortion, the positions of the CB and VB
were shied toward negative potentials, which facilitate better
hydrogen evolution.263
Carbon-rich carbon nitride polymer was also synthesized
using a hydrothermal route by heating melamine and glucose.
The –CHO and C–OH groups of glucose react with melamine to
give a more conjugated carbon extended network. Due to the
presence of a more conjugated network, the electrical conduc-
tivity and visible light absorption capacity were enhanced
Fig. 29 The atomic orbital diagram of the Frontier molecular orbitals and optimized geometry of C3N4 and F-doped C3N4 (CNF) having F atoms
attached in an out of plane fashion on corner carbon (CN–F1) and bay carbon (CN–F2). The CN–F2 has higher energy and displays more
prominent band shifting. Reproduced with permission from ref. 262 Copyright 2010 American Chemical Society.
View Article Online

(Fig. 30).264
Post calcination doping with carbon using glucose
has also been reported for hydrogen evolution (H2 production
rate of 40.37 mmol hÀ1
).265
In another study, Che et al. demon-
strated that the introduction of a carbon ring in between C3N4
framework (C ring)–C3N4 improved electron–hole pair separa-
tion due to an extended p network along with better visible light
absorption to achieve a high photocatalytic rate of hydrogen
production (371 mmol gÀ1
hÀ1
).266
Similar to the carbon ring
introduction in the carbon nitride framework, melamine was
also introduced into CN sheets by thermal annealing at high
temperature, resulting in the removal of some nitrogen and
formation of conjugated topological carbon nitride (TCN) with
Fig. 30 Schematic illustration of the polymerization process of melamine (MA) and glucose in a water solution to carbon-rich C3N4. Reproduced
and redrawn with permission from ref. 264 Copyright 2015 Royal Society of Chemistry.
Fig. 31 High resolution XPS of Cl-doped g-C3N4 (a) in the Cl2p region; inset shows the survey scan, (b) HR-XPS of g-C3N4 and Cl-doped g-C3N4
in the C1s region and (c) N1s region, and (d) crystal structure of pristine g-C3N4 and Cl-doped g-C3N4. Reproduced with permission from ref. 270
Copyright 2017 Elsevier.
View Article Online

a more extended conjugated network.267
In order to exploit the
band gap reduction via P doping and better charge mobility due
to an extended p-conjugated network, C- and P-co-doped
carbon nitride (CNPCN-1*) was synthesized using phytic acid
as a sole source of C and P.268
The resulting material exhibited
a much higher yield for hydrogen generation in photocatalytic
experiments (1493.3 mmol gÀ1
hÀ1
) which is ca. 9.7 times higher
than the performance of unmodied g-C3N4. Carbon nitride
materials modied with other heteroatoms, i.e., sulfur, have
also been reported for electrocatalytic and photocatalytic water
oxidation.269
Apart from doping via the replacement of carbon or covalent
functionalization, the strong intercalation of chlorine in the
carbon nitride framework was found to result in a better pho-
tocatalytic efficiency for hydrogen evolution (Fig. 31). Chlorine
intercalated g-C3N4 can be easily prepared by thermal treatment
of melamine and ammonium chloride. Mott–Schottky curves
revealed that the presence of chlorine upshied the positions of
the CB and VB. The electronic charge density plot showed that
charge density was concentrated on the Cl atom, indicating the
intercalation of Cl, which extends the local 2D p-conjugated
system of g-C3N4 into three dimensions and facilitates charge
separation (Fig. 32). The obtained Cl-modied g-C3N4 photo-
catalyst displayed 19.2 times more yield of H2 than pristine g-
C3N4, and an apparent quantum efficiency of 11.9% at 420 Æ
15 nm.270
By analogy Br-modied g-C3N4 exhibited a several-fold
increment in hydrogen production.271,272
8. Surface area modification of
carbon nitrides for improved hydrogen
evolution
Apart from doping, surface area and morphological manipula-
tion have been established to improve the adsorption of reac-
tants, making more active sites available for the reaction,
enabling the collection of more light and increasing the
resulting photoresponse. The surface area of carbon nitride
materials can be increased by the use of appropriate monomers
and their reaction under planned conditions or by templating
materials. Organic materials (especially surfactants such as
P123, Triton-X, CTAB, etc.) were investigated for so templating,
while inorganic silica was used as a hard templating material to
achieve a high surface area.273–276
For example, a mesoporous
carbon nitride structure prepared by a sol–gel method showed,
upon the removal of the silica template with HF, an improved
surface area and photoactivity for water splitting. High surface
area carbon nitride prepared with tetraethyl orthosilicate and
cyanamide (TEOS : CN) precursors in a 1 : 6 ratio was found to
give the best results (40.5 mL H2 in 24 h) under visible light
using a H2O/TEOA (9 : 1) mixture and Pt co-catalyst.277
In the
so template method of surface area improvement, the gener-
ated pores have nonspecic geometry and may collapse during
the heating step, which reduces the surface area. The use of an
ordered mesoporous silica structure as a hard template is
advantageous because in addition to an increase in the surface
area, the mesoporous geometry remains intact aer the removal
of silica. The Vinu group has done pioneering work on ordered
mesoporous carbon nitride synthesis by using SBA-15 as
a template.232
Numerous morphological modications in the
structure of carbon nitride have been achieved by this approach
(spherical, hollow spherical, tubes, 3D structures, etc.) by using
ordered silica sources, i.e., KIT-6, KCC and MCM-41.275,278,279
The
surface properties like the surface area of carbon nitride
materials can be further improved by using two or more carbon
nitride sources. Xu et al. synthesized nanoporous graphitic
carbon nitride having a 3.4 times higher surface area (46.4 m2
gÀ1
) by heating dicyandiamide and thiourea at a programmed
temperature.273
In this synthesis, thiourea served as a source of
the heptazine ring skeleton along with dicyandiamide and also
acted as a so template. Furthermore, the heating of specic
hydrogen bonded precursors can help in attaining a specic
morphology and crystallinity. Melamine and cyanuric acid form
a hydrogen bonded macromolecular structure in DMSO and
aer precipitation give rise to ower like spheres, which upon
annealing yield carbon nitride hollow spheres. These hollow
spheres displayed better absorption and lifetime with a small
broadening of the band gap. However, the morphology of these
spheres was distorted due to the evolution of ammonia during
the polycondensation step.280
In contrast, microwave thermol-
ysis of a melamine/cyanuric acid hydrogen bonded supramo-
lecular conjugate produced more crystalline, defect-free hollow
microspheres. A hydrogen evolution test in the presence of
triethanolamine (TEOA) and a Pt cocatalyst (5 wt%) gave 40.5
mmol hÀ1
, being two-fold higher than that of g-C3N4 (Fig. 33).281
The co-polycondensation of urea and oxamide produced g-C3N4
nanotubes (CN–OA). The CN–OA exhibited optical absorption
above 465 nm due to the n / p* transition of lone pairs of
nitrogen present on the edge of heptazine units. The apparent
quantum yield (AQY) of CN-OA-0.05 for H2 evolution with
a green LED (525 nm) reached 1.3%, which is approximately 10
times higher than that of bulk g-C3N4. Several nanostructured
morphologies have been recently prepared by utilizing the
melamine/cyanuric acid macromolecular structure under
various conditions.282
For example, Zhao et al. described the
synthesis of hollow mesoporous g-C3N4 mesospheres from
melamine cyanuric acid conjugates for water splitting.283
The
alteration of monomeric units was also accompanied by several
Fig. 32 Sketch of the visible light-assisted charge transfer and pho-
tocatalytic mechanism of Cl intercalated g-C3N4. Reproduced with
View Article Online

morphological transformations, i.e., 1,3,5-trichlorotriazine and
dicyandiamide aer hydrothermal treatment and thermal
annealing gave a spherical O-doped C3N4 photocatalyst.284
The
addition of sulfur powder (S8) was also found to promote the
synthesis of hollow vesicle shaped nanoparticles.285
Further-
more, microwave promoted synthesis of various morphologi-
cally modied g-C3N4 assemblies is gaining popularity due to
the fast reaction and efficient yield of products along with the
preservation of the structure. Defected holy g-C3N4 sheets were
prepared by microwave thermolysis of a hydrothermally diges-
ted dicyandiamide precursor for solar light mediated hydrogen
evolution (81.6 mmol hÀ1
).286
9. Carbon nitride sheets as
photocatalysts
The synthesis of carbon nitride through polycondensation of
organic nitrogen containing precursors afforded bulk agglom-
erated structures similar to graphite; bulk graphitic carbon
nitride possesses a layered structure in which tris-s-triazine
units are connected to each other via planar amino groups in
each layer. Therefore, bulk C3N4 can be delaminated into 2D
sheets which have a unique surface and photocatalytic proper-
ties. However, studies showed that similar to the isolation of GO
from graphite by Hummer's method, the synthesis of carbon
nitride sheets was difficult because of the strong covalent C–N
linkage between entangled sheets and hydrogen-bonding
between the strands of polymeric melon units with NH/NH2
group bonds.287
Attempts to synthesise carbon nitride sheets by
the Hummer's method resulted in the formation of g-C3N4
particles instead of nanosheets.288
However, other “top-down”
approaches such as thermal oxidation, etching, solvent
Fig. 33 Step (1): general outline of the synthesis of a hydrogen bonded melamine–cyanuric acid macromolecular structure in DMSO. Step (2):
thermal and microwave annealing of the macromolecular structure to hollow spheres. Upside SEM image of C3N4 hollow microspheres: (a) as-
synthesized and after thermal treatment at (b) 350
C, (c) 400
C, (d) 450
C, (e) 500
C, and (f) 550
C (reproduced with permission from ref. 280
Copyright 2013 Wiley-VCH). Low side (a, b) SEM images of CN540 and CN16; (c, d) TEM images of CN540 and CN16. The notation CN540
represents microspheres prepared by thermal treatment of the macromolecular structure at 540
C for 4 h and CN16 stands for microspheres
prepared by microwave treatment for 16 min. (Reproduced with permission from ref. 281 Copyright 2016 Wiley-VCH).
Fig. 34 Schematic illustration of solvent-assisted exfoliation of bulk g-
C3N4 to ultrathin nanosheets. Adapted with permission from ref. 289.
View Article Online

Sunlight-driven water-splitting using two dimensional carbon based semiconductors

Sunlight-driven water-splitting using two dimensional carbon based semiconductors

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Sunlight-driven water-splitting using two dimensional carbon based semiconductors

Similar to Sunlight-driven water-splitting using two dimensional carbon based semiconductors (20)

More from Pawan Kumar

More from Pawan Kumar (20)

Recently uploaded

Recently uploaded (20)

Sunlight-driven water-splitting using two dimensional carbon based semiconductors